Scattered evidence
Container restarts, logs, database locks and queues live in different tools. Investigation means switching context.
Ask your AI assistant to investigate the backend you already run. Backend Investigator connects evidence from your infrastructure so you can understand what happened, what it means, and what to check next.
Container restarts, logs, database locks and queues live in different tools. Investigation means switching context.
An error spike or a slow endpoint is visible, but determining why it happened still takes manual correlation.
Monitoring tells you something changed. You still need an explanation supported by observed facts.
Investigation results distinguish observations from hypotheses and identify missing signals. Specific answers depend on which sources you connect and what evidence is available.
Configure access to available sources such as Docker, PostgreSQL, Redis, BullMQ, Loki, Prometheus and GitHub.
Ask questions from a compatible AI client using MCP tools. The local Investigator gathers bounded read-only evidence.
Follow the evidence, inspect findings and decide what to change. The agent does not silently modify production infrastructure.
No automatic container restarts, SQL mutations or queue retries as part of investigation.
Work with existing infrastructure instead of forwarding all telemetry into a new platform.
Facts, hypotheses and source limitations are kept distinct to avoid confident but unsupported answers.
See the npm package for installation instructions, platform requirements, supported MCP clients and the current free-use license.
npx @mr.zedov/backend-investigator --help
Open npm package ↗
No account required for local operation. Local MCP usage telemetry is separate and opt-in; clicking an outbound link on this website is counted in aggregate.